Background Although noninvasive positive pressure ventilation (NPPV) has been successfully used for various kinds of acute respiratory failure, the data are limited regarding its application in postoperative respiratory failure after cardiac surgery. Therefore, we conducted a prospective randomized control study in a university surgical intensive care unit to evaluate the efficacy and safety of NPPV in the treatment of acute respiratory failure after cardiac surgery, and explore the predicting factors of NPPV failure. Methods From September 2011 to November 2012 patients with acute respiratory failure after cardiac surgery who had indication for the use of NPPV were randomly divided into a NPPV treatment group (NPPV group) and the conventional treatment group (control group). The between-group differences in the patients' baseline characteristics, re-intubation rate, tracheotomy rate, ventilator associated pneumonia (VAP) incidence, in-hospital mortality, mechanical ventilation time after enrollment (MV time), intensive care unit (ICU) and postoperative hospital stays were compared. The factors that predict NPPV failure were analyzed. Results During the study period, a total of 139 patients who had acute respiratory failure after cardiac surgery were recorded, and 95 of them met the inclusion criteria, which included 59 males and 36 females with a mean age of (61.5 +/- 11.2) years. Forty-three patients underwent coronary artery bypass grafting (CABG), 23 underwent valve surgery, 13 underwent CABG+valve surgery, 13 underwent major vascular surgery, and three underwent other surgeries. The NPPV group had 48 patients and the control group had 47 patients. In the NPPV group, the re-intubation rate was 18.8%, tracheotomy rate was 12.5%, VAP incidence was 0, and the in-hospital mortality was 18.8%, significantly lower than in the control group 80.9%, 29.8%, 17.0% and 38.3% respectively, P<0.05 or P<0.01. The MV time and ICU stay (expressed as the median (P25, P75)) were 18.0 (9.2, 35.0) hours and 4.0 (2.0, 5.0) days, which were significantly shorter than in the control group, 96.0 (26.0, 240.0) hours and 6.0 (4.0, 9.0) days respectively, P<0.05 or P<0.01. The postoperative hospital stays of the two groups were similar. The univariate analysis showed that the NPPV success subgroup had more patients with acute lung injury (ALI) (17 vs. 0, P=0.038), fewer patients with pneumonia (2 vs. 7, P<0.001) and lower acute physiology and chronic health evaluation II (APACHE II) scores (16.1 +/- 2.8 vs. 21.8 +/- 3.2, P<0.001). Multivariate analysis showed that pneumonia (P=0.027) and a high APACHE II score >20 (P=0.002) were the independent risk factors of NPPV failure. Conclusions We conclude that NPPV can be applied in selected patients with acute respiratory failure after cardiac surgery to reduce the need of re-intubation and improve clinical outcome as compared with conventional treatment. Pneumonia and a high APACHE II score >20 might be the independent risk factors of NPPV failure in this group of patients.